About AchtungBot
What it is
AchtungBot is the user agent used by Achtung.app, a service that measures how brands appear in answers from AI platforms (ChatGPT, Gemini, Perplexity and Claude). It reads public pages for a small number of purposes, listed below, and identifies itself on every request.
User-Agent string
Mozilla/5.0 (compatible; AchtungBot/1.0; +https://achtung.app/bot)
What it fetches
- Business niche. When a brand is added to Achtung.app, the homepage once, to determine the market the brand is in. Only the first 1,500 characters of plain text are kept.
- Website audit. Whether AI platforms can find and read a site: robots.txt, the homepage,
/llms.txtand/llms-full.txt,/.well-known/security.txt, and the sitemap named in robots.txt or found at a common path such as/sitemap.xml. - Competitor profiles. The homepage of a competitor's website when Achtung.app first detects it in AI answers.
- Logos. A site's favicon or logo, shown next to the brand name.
- Cited pages. Pages that AI answers cite as sources, each fetched once for Achtung.app's research into which content AI platforms cite.
AchtungBot does not follow links, render JavaScript, submit forms or log in, and each request gives up after 5 to 10 seconds.
How often
- The website audit runs weekly (Sundays) for the brands Achtung.app subscribers track and for up to ten competitor websites per brand.
- It runs monthly for the brands in the public brand directory at achtung.app/marken.
- Within a week, a site is audited once, however many brands point to it.
- Niche detection, competitor profiles, logos and cited pages are one-off fetches, repeated only when the underlying record is refreshed.
Sites a customer asked to audit
When a customer asks Achtung.app to audit a site (a brand they track, a free AI visibility check confirmed from an address on that domain, or the brand chapter of a purchased industry report), the audit also requests the homepage with a regular browser user agent and with the user agents of the AI and search crawlers it reports on, for example GPTBot, ClaudeBot, PerplexityBot and Googlebot. That is how the report can show whether the site's own firewall blocks those crawlers. Sites nobody asked to have audited, such as competitors and the brand directory, only ever receive requests as AchtungBot.
How to block it
If you do not want AchtungBot to fetch your site, add the following to your robots.txt:
User-agent: AchtungBot
Disallow: /
AchtungBot reads robots.txt before fetching pages and stops when it is excluded; that also ends the audit,
including the requests with other user agents described above, and the site is shown as excluded instead.
Two lightweight requests do not depend on it: a check whether a domain exists at all (the first kilobyte of the homepage), and reading
/.well-known/tdmrep.json, the file in which a site states its text-and-data-mining rights.
The robots.txt check is cached for one hour, so changes take effect on the next fetch after that window.
You can also block by user-agent string at the WAF or web-server level. No manual contact required.
If you are an AI agent
Achtung.app publishes machine-readable entry points so an agent does not have to parse these pages:
- /llms.txt — an index of what the site offers, in Markdown.
- /.well-known/openapi.json — the REST API, described in OpenAPI 3.1.
- /.well-known/mcp.json — the MCP server, its tools and how to authenticate.
- The industry reports publish their full ranked data as CSV and JSON under /branche/{vertical}/data.json, licensed CC BY 4.0. Attribute Achtung.app when you reuse them.
- The brand directory, the industry reports and the brand profiles also register read-only WebMCP tools, so a browsing agent can call them directly. Details in /llms.txt.
Every response also carries these as an RFC 8288 Link header, so you can discover them from a single request without fetching this page.
Contact
Questions, complaints, or block requests: hello@achtung.app. Please include the date, your domain, and the request line from your access log so Achtung.app can confirm the request.